Comment on Associating Schemas with XML documents 1.0 (First Edition) - extensibility

This is a comment on the editor's draft of Associating Schemas with XML 
documents 1.0 (First Edition) [1]. 

The draft says both of the following:

* A "phase" pseudo-atrribute MAY be supplied to control processing with 
Schematron

* Documents must not specify other pseudo-attributes on xml-model 
processing instructions.  (I.e. other than those explictly listed)

Taken together these choices seem a bit odd to me.  The support for 
"phase" suggests that it's appropriate to provide attributes that 
facilitate use with particular schema languages, but no indication is 
given of how similar support might be provided for other, and perhaps as 
yet unseen, languages in the future.   My suggestion would be that you 
choose one of the following two paths:

* Indicate that future versions of the specification may provide support 
for other pseudo-attributes, and specify how processors should behave when 
encountering an attribute that's not understood.

* Remove the prohibition on other pseudo-attributes, and again, specify 
how processors should behave when confronted with one that is not 
understood.

If you prefer not to specify explicit rules for unexpected attributes, you 
might at least say something like:  because no general rule is here 
provided for dealing with the case where an unsupported attribute is 
provided, interoperability can only be assured for those explicitly 
itemized here. 

Thank you.

Noah

[1] http://www.w3.org/XML/2010/01/xml-model/

--------------------------------------
Noah Mendelsohn 
IBM Corporation
One Rogers Street
Cambridge, MA 02142
1-617-693-4036
--------------------------------------

Received on Thursday, 4 February 2010 22:47:55 UTC